Intel X25-M SSD Shows Long-term Fragmentation Issue


mainstream_ssdsSaving up the $379 or so to get an 80GB Intel X25-M Solid State Disk drive? It’s great that you want to help the global economy with more spending, but you might want to hold up that purchase. PC Perspective has an extensive, longer-term review of the 2.5-inch SSD drive that’s worth a read. Just be sure you don’t have anything on the stove when you start reading; it took me a good 20-minutes to get through. Did I mention it was extensive?

Here’s the skinny from Ars: Intel’s (s INTC) wear-leveling algorithms and write combining approaches appear to cause nearly irreparable fragmentation. That bogs down performance over time on the MLC unit. I say “nearly” irreparable because you can fix the fragmentation issue: by completely wiping the drive. Yikes! How bad can performance get over time? Normal performance is a peppy 250MBps for reads and 80MBps for writes. Here’s a pic of the reads and writes over time:


That’s a huge performance hit. I highly recommend the PC Perspective article, even if you’re not in the market for Intel’s X25-M drive. And if you do plan to buy one, this is a must-read; Even if it is extensive.



There is a difference between the internal (wear levelling?) and external (file system) fragmentation. I believe the relationship between these is not direct. None of the Windows OSs currently are really optimized for SSDs, so file system fragmentation will occur anyway, and free space fragmentation will force random writes and cause slow down of the SSD over time.

Diskeeper who make file system defraggers have a new add-on to Diskeeper ’09 called Hyperfast that optimizes file system free space on SSDs and some other stuff. I noticed it in my copy of DK’09 pro, but never had a chance to test it since I don’t own an SSD. But the company’s benchmarks show a nice improvement on Apacer drives.

There is still much unknown/speculation/misinformation about long term SSD reliability and performance I feel. Things will be clearer in a year or two, when SSDs have been through extended periods of heavy usage.


Since all three test machines used Windows XP or Vista, I wonder how much a factor NTFS played in this. It was reported a few months ago that NTFS fragments free space on flash drives. Could possibly be compounding the problem with Intel’s wear-leveling algorithm. If so, the bright side of this story could be that the SSD optimization in Windows 7 will help. I’d also be interested in seeing this test done with Linux or OS X.

Comments are closed.